Comment 71 for bug 204996

@tzu (https://bugs.launchpad.net/ubuntu/+source/linux/+bug/204996/comments/64)

I do not have an SMB server nor Windows machine to connect to. As an alternative, I tried pulling 5 GB of data from the Hardy machine via ssh over wired ethernet. Worked just fine:

$ ssh 192.168.0.205 "dd if=/dev/zero bs=1M count=5000" | time -p od -bc
0000000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000
         \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0 \0
*
5000+0 records in
5000+0 records out
5242880000 bytes (5.2 GB) copied, 623.952 s, 8.4 MB/s
47040000000
real 625.92
user 80.06
sys 44.80

During that time, this is what vmstat looked like on the Hardy machine:

$ vmstat 1 10
procs -----------memory---------- ---swap-- -----io---- -system-- ----cpu----
 r b swpd free buff cache si so bi bo in cs us sy id wa
 3 0 27656 10972 1204 11552 99 97 194 99 4811 1045 9 11 75 5
 0 0 27656 10976 1204 11560 0 0 0 0 11281 2344 20 24 56 0
 1 0 27656 10936 1204 11560 0 0 0 0 11289 2555 16 28 56 0
 1 0 27656 10932 1212 11560 0 0 0 68 11216 2403 18 26 56 0
 1 0 27656 10880 1212 11560 0 0 0 0 11139 2254 15 27 58 0
 0 0 27656 10940 1212 11560 0 0 0 0 10749 2212 16 26 58 0
 0 0 27656 10952 1212 11560 0 0 0 0 11246 2254 15 27 58 0
 1 0 27656 10928 1212 11560 0 0 0 0 10942 2300 16 20 64 0
 1 0 27656 10916 1212 11560 0 0 0 0 11320 2282 18 22 60 0
 1 0 27656 10908 1212 11560 0 0 0 0 11237 2367 17 27 56 0

This machine is also running on 48 MB of RAM with a 600 MB swapfile. I accomplished this by adding this stanza to my /boot/grub/menu.lst file and making it the default:

title Ubuntu hardy (development branch), kernel 2.6.24-16-generic - test
root (hd0,0)
kernel /boot/vmlinuz-2.6.24-16-generic root=UUID=538f72a2-fbc0-454f-a058-32d1a0cae8b3 ro mem=48M
initrd /boot/initrd.img-2.6.24-16-generic
quiet

I also disabled gdm in runlevel 2:

$ ls -1 /etc/rc*.d/*gdm
/etc/rc0.d/K01gdm
/etc/rc1.d/K01gdm
/etc/rc2.d/K01gdm
/etc/rc3.d/S30gdm
/etc/rc4.d/S30gdm
/etc/rc5.d/S30gdm
/etc/rc6.d/K01gdm

I tested swap and heavy disk access by running this command:

$ time -p seq 1 10000000 | sort -S 100M > /dev/null
real 483.37
user 74.66
sys 6.37

and while it was running, vmstat looked like this:

$ vmstat 1 10
procs -----------memory---------- ---swap-- -----io---- -system-- ----cpu----
 r b swpd free buff cache si so bi bo in cs us sy id wa
 1 2 132108 1188 56 1896 394 298 480 316 4904 1149 11 11 64 14
 0 1 132108 1456 52 1808 1908 276 2024 276 322 294 1 2 0 97
 0 1 132108 1416 52 1784 1984 2152 1984 2152 339 321 3 2 0 95
 0 1 132108 1548 52 1604 2712 3360 2712 3360 357 459 4 1 0 95
 0 2 132108 1504 48 1468 1288 512 1320 512 250 314 2 2 0 96
 0 2 132120 1048 48 1000 3732 1008 3732 1008 423 483 8 0 0 92
 0 1 132164 1352 48 920 2148 1844 2148 1844 356 346 2 1 0 97
 0 1 132180 1336 48 852 2540 1560 2540 1560 366 413 3 2 0 95
 0 1 132180 1568 48 800 2096 1408 2096 1408 321 390 3 2 0 95
 0 1 132180 1720 48 800 2100 2040 2100 2040 300 375 2 0 0 98

Still looking for the test that breaks Hardy. Suggestions?

Regards,
- Robert